Group #3 - Checkpoint 2

1. Introduction

Our team explores MLB batting performance indicators using data from Baseball Savant.

Decision-maker / Context:

  • Baseball analysts and coaches interested in understanding player performance and team-level KPIs.

Research Question:

  • Which batting metrics are most predictive of overall offensive value, and how does player/team performance over time indicate future success?

Impact:

  • Insights could guide player evaluation, team decisions, and strategy planning.

2. Data Summary

Primary Datasets:

  • MLB team stats 2021-2025.csv

  • MLB player stats 2021-2025.csv

Data Metrics Glossary:

  1) Player-level Metrics

Abbreviation Full Name / Description
last_name Player’s last name
first_name Player’s first name
player_id MLB’s unique ID for the player, used for data merging
year Season year the stats come from
player_age Player’s age during that season
pa Plate appearances
hit Total hits recorded
k_percent Percentage of plate appearances that end in a strikeout
bb_percent Percentage of plate appearances that end in a walk
batting_avg Batting average: how often the player gets a hit per at-bat
slg_percent Slugging percentage: measures hitting power by counting total bases per at-bat
on_base_percent How often the player reaches base by any method (hit, walk, hit-by-pitch)
on_base_plus_slg OPS: combined measure of getting on base (OBP) and hitting for power (SLG)
isolated_power ISO: measures pure power by counting only extra-base hits
babip Batting average on balls in play: how often a ball put in play becomes a hit (excl. HR & strikeouts)
b_rbi Expected/adjusted RBI metric based on quality of contact
xba Expected batting average based on batted-ball quality
xslg Expected slugging percentage based on batted-ball quality
woba Weighted on-base average: advanced metric measuring total offensive value using run-values for each event
xwoba Expected wOBA based on contact quality
xobp Expected on-base percentage
xiso Expected ISO (expected power output)
wobacon wOBA on contact only (ignores strikeouts and walks)
xwobacon Expected wOBA on contact based on batted-ball data
exit_velocity_avg Average exit velocity (mph) of all balls the player hits

  2) Team-level Metrics:

Abbreviation Full Name / Description
Win% Team winning percentage over the season
SLG Slugging percentage: measures hitting power by total bases
OBP On-base percentage: how often the team reaches base
RBI Runs batted in: total runs produced by the team
ISO Isolated power: extra-base hitting ability
OPS On-base plus slugging: combined measure of offensive value

Data characteristics:

  • Mixed continuous metrics, percentages, and categorical identifiers (player, year).

Summary statistics:

  • Highlight top 10 performers in each dataset.
  • Player/team projected performance.
  • Most predictive metric of overall offensive value.

3. Data Analytics

3.1 Team Projections (2026–2028)

These plots show projected team-level KPIs for the next three seasons of top 10 teams based on linear regression with correlated predictors.

Top Teams Selected: Based on 2025 Win%

Graphs and Captions:

Projected Win Percentage (2026–2028)
Projected Win Percentage
Description: Projected win percentage for top MLB teams. Shows expected growth/decline in team performance over three years.

Projected SLG (2026–2028)
Projected SLG
Description: Shows projected slugging performance of top teams. Patterns indicate which teams may improve power hitting.

Projected OBP (2026–2028)
Projected OBP
Description: Team on-base percentage projection, reflecting plate discipline and consistency.

Projected RBI (2026–2028)
Projected RBI
Description: Estimated run production by team, linked to scoring potential.

Projected ISO (2026–2028)
Projected ISO
Description: Measures team isolated power; highlights teams likely to hit extra-base hits.

Projected OPS (2026–2028)
Projected OPS
Description: Combined on-base plus slugging metric, giving a broad measure of offensive efficiency.

3.2 Player Projections (2026–2028)

These plots show projected player-level KPIs for the top 10 players based on prior performance and correlated metrics.

Graphs and Captions:

Projected SLG for Top Players
Projected SLG
Description: Expected slugging trends for top players. Highlights consistency and potential breakout performers.

Projected OPS for Top Players
Projected OPS
Description: Combines OBP and SLG for a holistic view of player offensive output.

Projected ISO for Top Players
Projected ISO
Description: Player isolated power trends, indicating extra-base hitting capability.

Projected Batting Average for Top Players
Projected Batting Average
Description: Tracks expected hit rate per at-bat for top players.

Projected OBP for Top Players
Projected OBP
Description: Player on-base percentage projections, reflecting consistency and plate discipline.

Projected RBI for Top Players
Projected RBI
Description: Expected run production per player, linked to scoring potential.

Projected Exit Velocity for Top Players
Projected Exit Velocity
Description: Exit velocity trend predictions; higher values often correlate with power hitting.

3.3 Scatter Plots & Heatmaps (Quick Reference)

These plots show relationships between exit velocity and player hitting stats, and overall correlations among KPIs.

EV vs SLG:
EV vs SLG

EV vs RBI:
EV vs RBI

EV vs OPS:
EV vs OPS

EV vs OBP:
EV vs OBP

EV vs ISO:
EV vs ISO

EV Correlation Heatmap vs Hitting Stats:
EV Correlation Heatmap

Team Rankings Heatmap (2021–2025 averages):
Team Rankings Heatmap

4. Conclusion

Our analysis combined historical MLB player- and team-level statistics with regression-based projections to estimate future offensive performance for 2026–2028. By examining KPIs such as SLG, OBP, OPS, ISO, RBI, Batting Average, and Exit Velocity, we generated insights into which metrics are most indicative of future success.

Key Findings:

  • OPS and ISO consistently emerge as the strongest indicators of future offensive performance at both the team and player levels. Teams and players maintaining strong OPS trends also show positive projections in Win%, RBI, and SLG.

  • Exit Velocity has moderate correlation with power metrics (ISO ≈ 0.65, SLG ≈ 0.62), but does not strongly correlate with most KPIs. → This means EV is useful, but not a primary predictor in our dataset.

Team Projections (2026–2028):

  • Projected Win%, SLG, OBP, ISO, and OPS show identifiable upward/downward trends, helping estimate which teams may strengthen or decline offensively.

Player Projections (2026–2028):

  • Top hitters show differentiated paths in SLG, OPS, ISO, AVG, OBP, and Exit Velocity, helping identify consistent elite hitters versus potential breakout performers.

Heatmaps and Scatter Plots:

  • Correlation heatmaps confirm how hitting metrics relate to each other, supporting the regression choices used in projections. Scatter plots visually confirm the moderate relationships between EV and power statistics.

Limitations:

  • The model only uses batting statistics; defensive and situational factors are not included.

  • Projections are based on linear regression, which may not capture sudden changes (injuries, role changes, coaching changes, etc.).

Future Work:

  • Add park effects, defensive WAR, sprint speed, pitch-level statcast features, injury history, and multivariate models.

  • Use machine learning methods for improved forecasting accuracy.

5. Policy Recommendation

Hypothetical Decision:

  • Prioritize player acquisitions, contracts, and lineup planning based on projected offensive output.

Data-driven Recommendation:

  • Focus on players and teams with consistently high OPS, ISO, and OBP. Consider potential breakout performers in mid-tier metrics.

Benefits:

  • Quantitative and objective evaluation of offensive potential.

Risks:

  • Ignores defensive contributions and external factors; requires continuous monitoring of updated metrics.

Team Collaboration Notes

Challenges:

  • Learning Git workflow and branch management
  • Collaborating in code writing

Victories:

  • Created a common repo
  • Created project outline
  • Assigned roles
  • Learning GitHub Desktop workflow (branch management, pushing/pulling)
  • Agreed on main indicators & KPIs to analyze

Individual Contributions

Member Role Contribution
Matthew G Gonzalez Project Lead / Co Head Developer Data cleansing + Writing codes + visualization + findings and write-up
Jacob D Lamothe Code Editor/Checker + Video Editor + Presentation/Narration Lead Checks code for mistakes/redundancies + statistical validation + Edits video at the end of project
Rodolfo Lazaro Visualization Designer Tableau plots + checking visualizations
Samir Soliman Head Developer Import data + write codes + statistical Validation/Model Evaluation + findings and write-up

The next section will be deleted from our final report.

Current Steps for Checkpoint 2

  1. Complete KPI calculations in Python and update tables.
  2. Generate initial visualizations and save them in output/.
  3. Update README.md with insights from analysis.
  4. Commit and push updates to Group_3_Checkpoint_2.0 branch regularly.
  5. Ongoing amendment of the final report (README.md).

Progress on the 8 Major Tasks

  1. We selected Option 1 — Baseball Savant MLB performance indicators.

  2. Repository created.Team are collaborating on GitHub (Commit → Pull → Push workflow is currently used)

  3. Data downloaded from baseballsavant.mlb.com

  4. Python & Tableau are currently used for analysis and visualizations: Scatter Plots for relationships between variables, Line Plots to show performance trends over years, Boxplots to visualize distributions of key metrics & Histograms for frequency distributions of variables.

  5. The team is working on KPIs and forecast: On-Base Percentage (OBP), Slugging Percentage (SLG), Isolated power, Batting avg. & Exit Velocity

  6. This README.md is Checkpoint 2 deliverable.

  7. The recorded video will show results with narration.

  8. The final report will be displayed in the README.md file on the repo landing page.